Pollination, Floral Display, and the Ecological Correlates of Polyploidy

Jana C. Vamosi* • Simon J. Goring • Brad F. Kennedy • Rachel J. Mayberry • Clea M. Moray • Lisa A. Neame • Nicole D. Tunbridge • Elizabeth Elle

Department of Biological Sciences, Simon Fraser University, 8888 University Drive, Burnaby, BC, Canada Corresponding author : * [email protected] ABSTRACT Polyploidy is recognized as having a significant impact on plant evolution, with estimates of half of all vascular plants being of polyploid origin. It was first suggested in the 1940’s that polyploid species are concentrated in perennial herbs and in northern, temperate climates. Since then, there has been further attention concentrated on the ecological correlates of polyploidy, with an emerging interest in how polyploidization influences the pollination of a species. Here, we highlight recent work in pollination dynamics subsequent to polyploidization and gather information on differences in flower display characters between polyploids versus diploids. We posit that more thorough study of the effects of polyploidization on floral display is required, as this sub-field potentially provides keen insights into when and where polyploidy may be favoured via its effects on pollen delivery. Furthermore, our review of the evidence reveals that correlations between polyploidy and floral display in terms of flower size and flower number may influence perceived correlations between polyploidy and self-compatibility. ______

Keywords: floral display, macroecology, polyploidy, pollination, self-compatibility

CONTENTS

INTRODUCTION...... 1 MACROECOLOGY OF POLYPLOIDY, MATING SYSTEMS, AND POLLINATION...... 2 Mating system correlates ...... 2 Pollination syndrome...... 2 FLORAL DISPLAY CORRELATES OF POLYPLOIDY – THEORETICAL CONSIDERATIONS ...... 3 POLYPLOIDY AND POLLINATION– EMPIRICAL PATTERNS...... 3 Polyploidy and pollinator choice...... 4 Comparison of floral display between polyploid and diploid individuals...... 4 Cross-taxa comparisons of Rosaceae and Ranunculaceae ...... 5 Cross-taxa comparisons in the wind-pollinated genus, Plantago ...... 6 DISCUSSION AND FUTURE DIRECTIONS ...... 6 ACKNOWLEDGEMENTS ...... 7 REFERENCES...... 7 ______

INTRODUCTION on pollinator dynamics. Sexual system persists as one of the most important Polyploidization is arguably the most pervasive process in traits contextualizing the establishment and maintenance of plant evolution, with perhaps over half of all vascular plants polyploid lineages (Pannell et al. 2004). It is widely spec- being of polyploid origin (Grant 1971; Haufler 1987; de ulated that polyploidy should be strongly correlated with Bodt et al. 2005). Various aspects of polyploidy have oc- self-compatibility for two reasons. First, the process of gen- cupied plant systematists-evolutionists for decades (Steb- ome doubling may disrupt self-incompatibility mechanisms bins 1971, 1980; Soltis and Soltis 1993). Because poly- that are maintained in the diploid state (Miller and Venable ploidy is such a dominant feature of flowering plant gen- 2000), and polyploids should exhibit increased immunity to omes, it has been argued that polyploidy confers evoluti- the negative effects of inbreeding depression due to the pre- onary advantages to plant species and has provided oppor- sence of multiple gene copies (Lande and Schemske 1985; tunities for different ecological interactions (herbivory, pol- Otto and Whitton 2000). Although this latter point has been lination), which have resulted in an increase in biodiversity disputed (Busbice and Wilsie 1966; Bennett 1976), the ex- (Thompson et al. 1997; Otto and Whitton 2000; Thompson pectation becomes that transitions to polyploidy should of- et al. 2004; Thompson and Whitton 2006). Because pollina- ten be concurrent with transitions to increased self-compat- tion dynamics figure heavily into the diversification of an- ibility and higher selfing rates. Second, neopolyploids are giosperms as well (e.g. Perez et al. 2006; van der Niet et al. generally the minority cytotype in the population and thus 2006), it is perhaps surprising that the connection between outcrossers are more likely to encounter the gametes of their pollination and polyploidy has received so little direct at- diploid progenitors. This leads to a disadvantage because tention. Here, we review the ecological and mating system resulting offspring are often triploids, which tend to have correlates of polyploidy and how these correlates may in- reduced fertility compared to those of even ploidy. Self- fluence, and be influenced by, the effect of polyploidization compatible polyploids can avoid this minority cytotype dis-

Received: 15 March, 2007. Accepted: 19 April, 2007. Invited Review Functional Ecosystems and Communities 1(1), 1-9 ©2007 Global Science Books advantage by producing polyploid, self-fertilized offspring. ibility may nevertheless be more common in polyploid cyto- Therefore, polyploid cytotypes that arise within self- types or young polyploid species collectively (hereafter col- compatible lineages, or increase in self-compatibility upon lectively referred to as neopolyploids). polyploidization, should be more likely to persist than those The conflicting evidence over self-compatibility and that arise in self-incompatible lineages and maintain their polyploidy may also suggest another important factor at self-incompatibility. Early studies sup-ported the prediction play during polyploidization, likely a factor that is related to that polyploidy is correlated with self-compatibility (Crane both mating system and polyploidy. The pollination system and Lewis 1942; Hecht 1944; Lewis 1947; Williams 1951; is a strong candidate in this regard, since this is a factor of Bateman 1952). However, a recent, comprehensive global importance, with known relationship to both mating between-species analysis refutes the existence of a strong, system and fitness. Considering the longstanding interest in widespread association between self-compatibility and mating system and polyploidy, and the widely recognized ploidy level between species (Mable 2004). Nonetheless, relationships between mating systems, pollination mecha- this result is not inconsistent with the theoretical prediction nisms, and floral display traits, there has been a surprising described above, since between-species comparisons do not lack of attention paid to floral display and the pollination of necessarily reflect transient trait values immediately prior polyploid cytotypes and species. Self-compatibility is ex- to or following polyploidization. Self-compatibility could pected to be common in neopolyploids; but additionally, be selected in neopolyploids due to reduced minority cyto- selfing species often have relatively small, inconspicuous type disadvantage and reduced inbreeding depression; but flowers and low pollen-to-ovule ratios (Cruden and Lyon as substantial premating reproductive isolation builds up 1985; Cruden 2000). If polyploidization is strongly correl- between diploids and polyploids, perhaps facilitated by ated with self-compatibility then neopolyploid species functional divergence between duplicate gene copies, the should generally have very small flowers as well. Anecdotal selection for strong self-compatibility could decrease or evidence thus far indicates that polyploid species are found reverse. Likewise, although polyploidization may in some among lineages spanning a large range of flower sizes, from cases cause a disruption of self-incompatibility mechanisms, small wind pollinated grasses (of which, the majority are this does not preclude the possibility that self-incompatibi- polyploid) (Hilu 2004), to the large, insect pollinated inflor- lity could be reestablished over the longer term, especially escences of Chamerion angustifolium (Husband and Sabara when self-compatibility is intermediate and/or variable 2004; Fig. 1). This indicates that the relationship between between genotypes (Mable 2004). Therefore, although there pollination mechanism, self-compatibility, and polyploidy is is at best only weak evidence that self-compatibility is more not amenable to interpretation without information regar- common in contemporary polyploid species, self-compat- ding its phylogenetic context. We intend this review to (1) point out some pollination-related macroecological factors that may be confounded with previously-documented correlates of polyploidy; (2) compile existing data regarding floral display and polyploidy; and, (3) indicate future research that would help to elucidate which pollination-related factors facilitate polyploid establishment.

MACROECOLOGY OF POLYPLOIDY, MATING SYSTEMS, AND POLLINATION

Mating system correlates

Mating system, like polyploidy, is not uniformly distributed with regard to latitude, altitude, and insularity (Richards 1997). The often-reported correlation between self-compatibility and polyploidy may actually be the result of under- lying association between mating system and geography and/or habitat type. Polyploidy has been associated with greater incidence at higher latitudes (Love and Love 1943; Haskell 1952) and altitudes (Johnson and Packer 1965; Petit and Thompson 1999; Brochmann et al. 2004; Guggisberg et al. 2006), as well as in island floras, including that of New Zealand (Hair 1966), Norfolk Island (De Lange and Murray 2003), and Hawaii (Carr 1978). Self-compatible species are expected to be better able to colonize islands (Baker 1959) while tropical floras are thought to exhibit higher degrees of self-incompatibility (Bawa 1990). Self compatibility is as- sumed to be important at high altitudes (but see Smith and Young 1987) because of harsh conditions and short growing seasons, and some degree of self compatibility has been de- monstrated at many high latitudes (e.g., tundra plants; Mo- lau 1993). These patterns are consistent with the idea that polyploidy is highest where self-compatibility is highest. It is currently uncertain whether the general geographic pattern of polyploidy mirrors that of self-compatibility due to a direct association between polyploidy and self-compatibility or because these life history traits are independently favoured in common habitats.

Pollination syndrome

The pollination ecology of most polyploid species is scar- cely known, but tentative predictions can be made based on Fig. 1 Diploid (left) and tetraploid (right) Chamerion angustifolium. the type of pollinator communities that are found in areas Photo courtesy of Brian Husband. with high occurrence of polyploidy. Island habitats differ in

2 Pollination ecology of polyploidy. Vamosi et al. their suites of pollinators, and are described variously as Polyploids are expected to differ in flower size from rich in small bees (New Zealand; Newstrom and Robertson their diploid progenitors (Stebbins 1971), yet whether they 2005), nectarivorous birds (Macronesia; Valido et al. display an overall bias towards larger or smaller flowers is 2004), and dipterans (Juan Fernadez islands; Bernadello et not clear. Flower size is expected to diminish in polyploids al. 2001). In general, however, island flora are thought to relative to diploid relatives due to the previously discussed have small, simple flowers and more generalist pollinators correlations between mating system, geography, and ploidy. (Olesen et al. 2002; Abe 2006). Arctic and alpine regions In contrast, polyploids can have larger flowers than their show a dominance in flies as pollinators (Yumoto 1986; diploid progenitors due to the “gigas effect”, which is an Elberling and Olesen 1999) and fly-pollinated species are increase in polyploid cell size that often results in increased also more often self-compatible (Yumoto 1986). Converse- organ size throughout the plant (Stebbins 1971). Although ly, moist tropical rainforests, which are thought to have these increases in cell size and growth appear to quickly relatively low frequencies of polyploidy (Love and Love decrease within a few generations (Otto and Whitton 2000), 1943; Morawetz 1986), are thought to also have a high their rapid but short-lived manifestation may still be of incidence of pollinator specialization (Olesen and Jordano great importance during the initial establishment phase 2002). This information suggests that the context in which when polyploid cytotypes are rare. Larger floral displays of polyploidization is thought to occur is most frequently polyploids may encourage visitation by pollinators (Taylor upon a background of self-pollination (or at least self-pol- and Smith 1979; Totland 2001), or even change the relative len deposition) via generalist pollinators. Wind pollination frequency by which specialist versus generalist pollinators also increases with altitude and latitude, and is relatively visit. However, selection for larger floral displays may only uncommon in the tropics (Regal 1982), suggesting that be important when pollinators are choosy, scarce, or unpre- anemophily may also provide a context for polyploidiza- dictable (Vamosi and Otto 2002; Harder and Johnson 2005), tion. increasing the likelihood of floral display influencing poly- Based on the macroecology of polyploidy, mating sys- ploid establishment in alpine and tundra conditions. tem, and pollination syndromes, we expect that either in- The combination of theoretical expectations from mo- sect pollination by generalist small bees and/or flies, or dels including inbreeding depression and our understanding wind pollination, will predominate for the majority of of pollination biology of polyploids is complex. We can polyploid species. We can further speculate on the expec- predict small inflorescences (to reduce pollen discounting) ted floral phenotypes of entomophilous polyploids if we and small flowers (to increase selfing and so reduce the fre- consider the typical morphological traits of plant species quency of mating with diploid progenitors), but incidental with generalist pollinators and high selfing rates. Plant increases in flower size with doubling of the chromosome species that are typically visited by generalist pollinators content can increase pollinator attraction as well as herkog- are thought to be radially symmetric (e.g., Sargent 2004). amy, thereby decreasing the selfing rate (Webb and Lloyd Studies within specific groups, such as Euphrasia (French 1986; Barrett and Eckert 1990). If these morphological and et al. 2005) and global cross-species comparison (Snell and mating system changes promote outcrossing, then it would Aarssen 2005) indicate that selfing species often have seemingly amplify the effects of minority cytotype disad- smaller flowers. Mechanistically, this correlation between vantage when polyploid cytotypes are rare, and thereby re- small flowers and selfing may arise due to differences in duce the likelihood of tetraploid establishment. This could development, with large-flowered species experiencing be counteracted if assortative mating takes place, either via delayed anther-stigma contact compared to small-flowered pollinator constancy to cytotypes that demonstrate variation sister taxa (Armbruster et al. 2002). Finally, some prelimi- in floral rewards, or via deposition of pollen on different nary evidence indicates that selfing species have more locations of a pollinator’s body due to floral-size differen- flowers per inflorescence (Sato and Yahara 1999), than do ces between cytotypes (the latter mechanism perhaps being relatively more outcrossing species. more likely in bilateral flowers). While polyploidization can create initial differences in floral morphology amenable FLORAL DISPLAY CORRELATES OF to assortative mating, subsequent evolutionary reinforce- POLYPLOIDY – THEORETICAL CONSIDERATIONS ment could be expected to promote the evolution of further reproductive isolation due to the fitness cost of producing The interaction between selfing rates, inbreeding depres- mixed-ploidy offspring (Whitton 2004; Nuismer and Cun- sion and autotetraploid establishment has been recently ex- ningham 2005). Such isolating mechanisms need not be plored in more detail with two models (Rausch and Mor- qualitatively the same as those that are directly tied to poly- gan 2005; Yamauchi 2006). Rauch and Morgan (2005) de- ploidization, and the extra genomic content of a polyploid monstrated that higher selfing rates and lower inbreeding may facilitate relatively rapid evolution of novel floral phe- depression in autotetraploids are expected to facilitate es- notypes. For example, over longer time periods, the ad- tablishment of autotetraploid populations. An important ditional genomic content within polyploids may facilitate outcome of these models is the identification of pollen lim- the evolution of novel structures such as nectar spurs and itation (Rausch and Morgan 2006) and pollen discounting additional petals (Levin 1983). Below, we review the evi- (Yamauchi 2006) as important factors in autotetraploid es- dence of the correlates of polyploidy with floral display tablishment. In order to better understand the implications traits of plants and reflect further on how these correlates of these models on geographic and phylogenetic patterns of arise. polyploidy we must consider the relationship between pollen limitation and flower size (Knight et al. 2005) as well POLYPLOIDY AND POLLINATION– EMPIRICAL as the ecological determinants of pollen limitation. We PATTERNS must also determine whether changes in floral morphologies with polyploidization tend to result in higher self- There is surprisingly little direct empirical evidence on the fertilization (e.g., reduced flower size and pollinator re- relationship between floral display and pollinator preferen- ward) (Schoen et al. 1996). In Yamauchi’s model (2006), ces in mixed cytotype populations despite the recognition increases in pollen discounting make autotetraploid estab- that the extent of pollen transfer within and between newly lishment more difficult. Pollen discounting is thought to formed polyploids, triploids, and diploids represents an im- increase when species have large inflorescences (Harder et portant first step (or barrier) in tetraploid establishment al. (2004) and references therein) and a large proportion of (Thompson and Lumaret 1992; Thompson et al. 1997, self pollen is delivered to flowers on the same plant 2004). However, some inferences can be made by indirect through within-plant pollinator movement (geitonogamy). comparisons of floral morphologies and pollinator syn- Thus, if the Yamauchi model has widespread applicability, dromes. Larger polyploid flowers are evident in Chamerion polyploidy should be rare amongst species with large angustifolium (Husband and Schemske 2000; Kennedy et al. (many-flowered) inflorescences. 2006). Yet there are numerous cases (e.g., Claytonia parvi-

3 Functional Ecosystems and Communities 1(1), 1-9 ©2007 Global Science Books flora (Miller and Chambers 1993); Tarasa (Tate and Simp- discriminate between cytotypes. More studies are clearly son 2004), and Amsinckia gloriosa (Johnston and Schoen needed to address the long-standing hypothesis that poly- 1996)) in which polyploid flowers are smaller and have ploid establishment should be favored when there is high higher selfing rates than diploid progenitors. We review (1) flower constancy of pollinators. If flower constancy is im- the studies that have directly examined pollinator dynamics portant for polyploid establishment, we would predict the between polyploid and diploid pollinator dynamics; 2) the incidence of polyploidy should be higher in plant species studies that have examined floral display metrics between pollinated by social bees or birds compared to those that are polyploid and diploid individuals of the same or closely fly or wind pollinated. Yet, this prediction is opposite to related species. We divided autopolyploids and allopoly- that expected from the higher levels of polyploidy in eco- ploids because widely disparate mechanisms are required systems dominated by generalist pollinators (i.e. high alti- to generate polyploids in each. Because there may be a tudes or latitudes). To our knowledge no comparative test research/publication bias, in that cases where polyploids of these predictions has been conducted. and diploids differ in floral features are more likely to be examined, we have also gathered floral display metrics for Comparison of floral display between polyploid species within Ranunculaceae, and the genera of Rosaceae and diploid individuals and Plantago, where ploidy information was available to investigate whether there are predictable cross-taxa cor- We compiled the evidence for increases in floral display relates between polyploidy and flower size, the number of with increases in ploidy level for autopolyploids (Table 1). flowers per inflorescence, and pollination syndrome at- We performed a literature search in both the Web of Sci- tributes. ence and Google Scholar for combinations of the key words polyploid* and flower (or floral) morph* or size. We also Polyploidy and pollinator choice worked backward through all references identified to locate additional relevant studies. This resulted in 18 species for What little evidence is available seems to indicate that pol- which the size and/or numbers of flowers was compared linators with a high degree of flower constancy may facil- between diploids and autopolyploids. We found that poly- itate polyploid establishment. A high degree of flower ploid individuals had larger flowers than their diploid coun- constancy increases the chance that newly formed poly- terparts more often than expected by chance (14 of 17 spe- ploid genotypes will receive pollen from plants of like cies; P = 0.01; sign test). Flower number was reported less ploidy and avoid the strong fitness costs associated with frequently, and was larger for polyploids in just 2 of 5 stud- producing unfit hybrids (Levin 1975). Kennedy et al. ies (Table 1). (2006) found that within mixed-ploidy populations of C. This literature search also identified 13 studies that angustifolium, 73% of all pollen came from within-ploidy compared flower size or number between a polyploid spe- pollinations, mirroring pollinator foraging behaviour. Tet- cies (or group of polyploid species) and either closely rel- raploid establishment is likely facilitated in mixed-ploidy ated species or species known or believed to be the progen- populations of C. angustifolium because pollinators also itors of an allopolyploid. In this data set, which compares exhibited higher constancy on tetraploids and visited them species that have diverged over a longer evolutionary time disproportionately, likely because tetraploids have signifi- frame than those in the autopolyploid data set, we find that cantly larger and more open flowers per inflorescence than polyploids had larger flowers than diploid counterparts in diploids (Husband and Schemske 2000; Kennedy et al. just 6 of 13 cases (P >0.05; sign test), and in only one of 2006). Similarly, floral visitors strongly discriminated the three studies examining flower number did polyploid between diploid and tetraploid Heuchera grossularifolia, individuals have more (Table 2). which also exhibits morphological differences between cy- These results highlight two issues. First, flower size totypes (Segraves and Thompson 1999). Although visita- may not be universally larger in polyploids relative to dip- tion rate did not differ between the cytotypes in H. gros- loids, despite the assumption that this is a common out- sularifolia, diploids were visited preferentially by sweat come due to the gigas effect. Autopolyploids do tend to bees and bumble bee workers, while tetraploids were pre- have larger flowers than diploids within the same species, ferentially visited by beeflies (Bombyllius major), moths and this is the case in the two species that are well-studied (Greya politella) and bumble bee queens (Bombus spp.) in terms of their pollination biology (C. angustifolium and (Segraves and Thompson 1999). H. grossularifolia). Flower size differences could assist These two species are apparently the only ones that polyploid establishment in this instance, as discussed previ- have been examined to test the idea that pollinators might ously. Secondly, it is unclear how much publication bias

Table 1 Within-species comparisons of flower display between polyploid and diploid species for autopolyploids. Species Flower size Larger?1 Flower number Larger? Reference(s) Anemone palmata Tepal number, length P Medail et al. 2002 Alyssum maritimum Petal length P Bali and Tandon 1959 Antirrhinum majus Flower size P Emsweller and Ruttle 1941 Arrhenatherum elatius Panicle length P Petit et al. 1997 Asphodelus fistulosis Tepal length D Ruiz Rejon et al. 1990 Chamerion angustifolium Flower width P # Open flowers P Husband and Schemske 2000; Kennedy et al. 2006 Dactylis glomerata Panicle length P Lumaret et al. 1987 Deschampsia cespitosa Glume size P Rothera and Davy 1986 Heuchera grossularifolia Petal length,width P # Open flowers D Segraves and Thompson 1999 Hibiscus syriacus cultivars Corolla diameter P van Huylenbroeck et al. 2000 Lotus alpinus Petal size P Flowers per umbel P Gauthier et al. 1998 Lilium formasium Flower size P Emsweller and Ruttle 1941 Lycium californicum Flower length P Yeung et al. 2005; J. Kohn pers comm Phlox drummondii Inflorescence diameter P Flower number D Garbutt and Bazzaz 1983 Ranunculus adoneus Floral morphology I Baack 2005 Sedum pulchellum Flowers per inflorescence D Smith 1946 Tradescantia spp. Floral morphology I Anderson and Sax 1936 Zea mays Flower size P Randolph 1935 1P = polyploidy species is larger; D = diploid species is larger; I = polyploids and diploids are indistinguishable, or polyploid is intermediate

4 Pollination ecology of polyploidy. Vamosi et al.

Table 2 Among-species comparisons of flower display between polyploid and diploid species for allopolyploids. Polyploid species Diploid Species Flower size Larger?1 Flower number Larger? Reference(s) Allium cepa x fistulosum Allium cepa, A. fistulosum Flower size P Levan 1941 Amsinckia gloriosa Amsinckia douglasiana Floral length, width D Johnston and Schoen 1996 Anthericum liliago Anthericum ramosum Flower diameter P Flower number D Rosquist and Prentice 2001 Claytonia parviflora utahensis Claytonia parviflora Flower size D Miller and Chambers grandiflora 1993 Collinsia parviflora Collinsia parviflora, Flower size I Tunbridge and Elle C. grandiflora unpublished Digitalis mertonensis D. purpurea, D. ambigua Flower size P Buxton and Darlington 1931 Draba spp. Draba spp. Petal area P Brochmann 1993 Euphrasia minima Euphrasia christii, Corolla length I Liebst and Schneller E. hirtella 2005 Platanthera huronensis Platanthera aquilonis, "floral features" I Wallace 2004 P. dilatata Senecio cambrensis Senecio squalidus, Capitulum length P Abbott and Lowe 2004 S. vulgaris Stellaria longipes Stellaria longifolia Flower diameter P Flowers per ramet D Macdonald and Chinnappa 1988 Tarasa spp. Tarasa spp. Petal length D Tate and Simpson 2004 Tragopogon miris, T. miscellus Tragopogon dubius, Not examined Flowers per head P Ownbey 1950 T. porrifolius, T. pratensis Vaccinium uligunosum Vaccinium myrtillus, Corolla diameter D Jacquemart and V. vitis-idaea Thompson 1996 1P = polyploidy species is larger; D = diploid species is larger; I = polyploids and diploids are indistinguishable, or polyploid is intermediate. influences this result; in some cases the absence of mor- Cross-taxa comparisons of Rosaceae and phological differences from papers with a different focus Ranunculaceae (e.g. Ranunculus adoneus, Baack 2005) and in one case not included in Table 1 we learned of a lack of morphological This review suggests that autopolyploid species and genera difference by inquiring of the author (for Hieracium echi- should have larger inflorescences than diploid species and oides, T. Peckert pers. comm.). genera, though the case is less clear for allopolyploids. We The lower incidence of larger flowers in allopolyploids delve further into the floral traits (flower size, floral syn- relative to autopolyploids may be due to the differing ge- drome, flower color) of polyploids versus diploids within netic mechanisms involved in hybridization versus geno- Ranunculaceae and Rosaceae. mic doubling. It has been argued that selfing especially We obtained ploidy information for Ranunculaceae facilitates polyploidization in allopolyploids (Grant 1971), species included in Mable (2004) from Barbara Mable, and attributable to there being a high frequency of spontaneous also used the proportion of polyploid species per genera of tetraploids produced by selfed hybrids (Ramsey and Rosaceae from Vamosi and Dickinson (2006). For the spe- Schemske 1998). This is thought to be responsible for the cies and genera within these datasets we found coarse esti- pattern of self-fertilizing polyploids being most commonly mates of mean flower diameter (or mean flower depth in of allopolyploid origin (Stebbins 1957). Thus, pollinator the case of zygomorphic flowers) and flowers per inflores- dynamics may be relatively unimportant in allopolyploid cence (coded as solitary vs. not solitary) in species descrip- species and the larger flowers observed in (relatively more tions. The main references for these data include (Kalkman outcrossing) autopolyploids may reflect selection for in- 2004; Tamura 2004), and the E-flora database (www. creased pollination. Alternatively, recent molecular work efloras.org). For Ranunculaceae, we also coded flower co- in neopolyploid species indicates that the rate of gene si- lor, and flower symmetry traits, as these are expected to in- lencing between duplicated allopolyploid and autopoly- fluence pollinator choices. Too little variation occurs in ploid genomes may differ substantially (Adams and Wen- Rosaceae to analyze these traits (most Rosaceous species del 2005) and this may play a role in whether the gigas ef- have actinomorphic, white/pink flowers). fect is observed. Clearly, further research comparing mor- We find that these coarse floral display metrics reveal phology of polyploids with their progenitors is needed to important correlations between floral display and poly- address whether our long-standing assumption about rel- ploidy. In Ranunculaceae, polyploidy was surprisingly ative flower sizes is true. Too few studies are published at equally proportioned within species in terms of symmetry this time for strong conclusions to be made, which is why and flower color (Table 3). However, polyploid species we chose to improve the number of species within our had, on average, smaller flowers than diploid species and comparison by compiling data for cross-taxa comparisons. more often displayed multi-flowered inflorescences. In

Table 3 Summary of floral display comparisons between species in Ranunculaceae and genera of Rosaceae. Ranunculaceae Rosaceae Trait N Direction of correlation with polyploidy N Direction of correlation with polyploidy (P-value) (P-value) Flower size (mm) 89 – (0.013)1 71 Ns (0.171)2 More than one flower per inflorescence 117 + (0.026) 80 + (0.010) (vs. solitary flowers) Symmetry (actinomorphic vs. zygomorphic) 120 Ns (0.589) N/A Not done Color (blue/red/purple vs. 108 Ns (0.186) N/A Not done white/green/cream/yellow) 1 P-values from contingency tests (or t-tests in the case of flower size) between species that were 1-diploid; 2-polyploid; or 3-mixed. Ns = not significant. 2 P-values from t-tests (for solitary flowers or not) or correlation tests (flower size), comparing arc-sin proportions of polyploid species per genus in Rosaceae. Flower size was ln-transformed mean (in mm) of values obtained for the species or genus. Ns = not significant.

Rosaceae, there is no significant correlation between prop- Chasmogamous neopolyploids may then be at an initial ortion of polyploidy and mean flower size (Table 3). How- selective disadvantage because gigas effects would appear ever, there is strong pattern for genera with solitary flowers to limit pollen capture (changes in corolla size) and pollen to have a much lower incidence of polyploid than genera dispersal (changes in pollen size). These effects may be with >1 flowers/inflorescence (F1,78 = 6.99; P = 0.010). Be- ameliorated through rapid selection either for smaller pol- cause of the strong flower size-number trade-off that that is len grain and corolla sizes or by transitioning to a cleistoga- observed in many species (Cohen and Dukas 1990; Harder mous mating system. Determining how many independent and Barrett 1995; Worley et al. 2000), it is perhaps not sur- evolutionary transitions to polyploidy in chasmogamous prising that a cross-taxa comparison does not reveal flower and cleistogamous lineages would be particularly instru- size to be larger for genera with high levels of polyploidy. mental in revealing whether gigas effects impose any level of constraint on the evolution of polyploidy in chasmoga- Cross-taxa comparisons in the wind-pollinated mous lineages, but this must await phylogenetic analysis. genus, Plantago Polyploidy has effects on both entomophilous and anemophilous pollination syndromes. A large body of informa- Because anemophily (aerial pollination) is known to be im- tion exists for plants with abiotic pollination schemes that portant in geographic areas where polyploidy is common, would shed more light on the interactions between mating we were interested in examining morphological traits in systems and polyploidy. wind-pollinated species that had both high rates of polyploidy and variation in mating system (cleistogamy, or DISCUSSION AND FUTURE DIRECTIONS obligate selfing, and chasmogamy, capable of outcrossing). Wind-pollinated species generally have small pollen grains Geographical and temporal variation in pollinators is recog- capable of being transported long distances (Ackerman nized as a complicating factor in the evolution of floral dis- 2000) and floral parts that modify airflow around repro- play (Sahley 1996; Buide 2006) but this relationship has re- ductive structures to maximize pollen capture. Evidence ceived little attention in its effects of polyploid origin and for the adaptive value of pollen grain size comes from a maintenance. In any study of polyploid evolution, a basic study of Plantago demonstrating that pollen grains that ar- problem is disentangling the initial effects of polyploid for- rive on stigmas are smaller than those produced by the spe- mation from the divergence occurring subsequent to poly- cies as a whole (Primack 1978) indicating strong selection ploid formation (Stebbins 1971, 1980), and this issue has pressure for this trait. If the gigas effect is a widespread been considered in recent reviews (Ramsey and Schemske phenomenon, anemophilous polyploids should display lar- 1998; Wendel 2000; Ramsey and Schemske 2002; Osborn ger pollen grain size and larger corolla size unless these 2004). Because our study was performed at both the within- traits are under strong selective pressure, either for pollen species and between-species levels, it gives some insight transport (grain size) or pollen capture (corolla size). into the processes that might result in the patterns we see. We compiled data on the genus Plantago, for which in- Because within-species comparisons indicate that polyploid formation was available on mating system and pollen size individuals most often have larger flowers than their diploid (Bassett and Crompton 1968) as well as chromosome num- counterparts, polyploidization appears to most often result bers (Bassett and Crompton 1968; Rahn 1996). For the in larger flowers. However, theoretical predictions sugges- purpose of this review, we define polyploidy in Plantago ting that polyploidization should occur most often in species as any 2n chromosome count greater than 12 (Stebbins with (many) small flowers also appear to be true. Thus, spe- 1971). For the Plantago species examined, polyploidy is cies where only polyploid cytotypes are known (or species more common than the original diploid state (Table 4) where both polyploid and diploid races are known) have while the number of cleistogamous species is roughly smaller flowers than purely diploid species or genera be- equal to the number of chasmogamous species. cause of flower number/size tradeoff and we find correla- Polyploidy was not significantly associated with cleis- tions between small flowers and polyploidy even though togamy or chasmogamy (2= 0.2015, P >0.6535). Pollen polyploidization may increase flower size. grain size is non-normally distributed and is larger in poly- To our knowledge this paper presents the first report that ploids than diploids (Wilcoxon test, W=158.5, P <0.01). polyploid species typically have more than one flower per Polyploids also had larger corollas than diploids (t-test, t = inflorescence. This result was found in both Rosaceae and 3.9, df = 13.115, P <0.01). Pollen sizes were not signifi- Ranunculaceae indicating that this may be a general obser- cantly different between cleistogamous and chasmogamous vation and not one restricted to a single family. The few stu- species. Cleistogamous species have more variable pollen dies that have measured self-pollination within and among grain sizes than chasmogamous species although there is flowers found that geitonogamy contributed over 40% of no significant difference in size. Cleistogamous polyploid the self-pollination within inflorescences (Leclerc-Potvin pollen grains are significantly larger than cleistogamous and Ritland 1994; Schoen et al. 1996; Eckert 2000; Karron diploids (Wilcoxon test, W = 33, P <0.01). This pattern is et al. 2004). Interestingly, our findings of increased flower also seen in corolla size (t-test, t = 9.8445, df = 8, P number provides another avenue leading to a correlation be- <0.001). tween polyploidy and increased selfing because the inci- Chasmogamous species have smaller pollen grain sizes dence of geitonogamy generally increases with display size and smaller corolla sizes regardless of chromosome com- (Harder and Barrett 1995; Brunet and Eckert 1998; Rade- pliment. These results support the hypothesis that strong maker et al. 1999; Karron et al. 2004), as pollinators visit selection pressures on chasmogamous species to maximize more flowers per inflorescence. Geitonogamy may thus pro- pollen transport and capture operate to reduce pollen size vide a mechanism for increasing rates of polyploidy by pro- in these species. The presence of a gigas effect would then viding a level of selfing necessary for the establishment of a only be seen in cleistogamous species where we do see dif- polyploidy population within a diploid matrix. Certainly this ferences in both corolla size and pollen grain size between mechanism ignores the very real possibility of geitonogamy polyploids and diploids since the variation as a result of among multiple single flowers on a plant, and we simply polyploidy can be maintained without any direct cost to wish to point out that the connection between geitonogamy reproductive success. and polyploidy provides fertile ground for future study. Further study is also required to tease apart the con- founding influences that still mar our ability to detect which Table 4 Two way table of species properties for Plantago. traits facilitate polyploidy establishment. First, if genomic Diploid Polyploid Total doubling results in an increase in the number of flowers per Chasmogamous 6 11 17 inflorescence (Kennedy et al. 2006), then we can not infer Cleistogamous 3 11 14 that polyploids establish more readily in many-flowered lin- Total 9 22 31 eages. However, we see little evidence that this is a general

6 Pollination ecology of polyploidy. Vamosi et al. consequence of genomic doubling (see Tables 1, 2). Sec- ACKNOWLEDGEMENTS ond, because polyploidization can disrupt self-incompatibility mechanisms, it is possible that polyploidy establishes We thank B. Mable for generously providing her dataset and B. within self-incompatible species with solitary flowers and Husband for providing the photo of Chamerion angustifolium. differences in selfing ability and/ or the number of flowers per inflorescence evolves secondarily. It is thought that REFERENCES self-incompatibility mechanisms break down because of several genetic factors, including increased heterozygosity Abbott RJ, Lowe AJ (2004) Origins, establishment and evolution of new poly- of the alleles that control pollen phenotype or the loss of ploid species: Senecio cambrensis and S. eboracensis in the British Isles. Bio- function of RNases that break down self-pollen, and dif- logical Journal of the Linnean Society 82, 467-474 ferent mechanisms may be involved in sporophytic versus Abe T (2006) Threatened pollination systems in native flora of the Ogasawara gametophytic incompatibility (Chawla et al. 1997; de Net- (Bonin) Islands. Annals of Botany 98, 317-334 tancourt 2001). What is not known, however, is whether the Ackerman JD (2000) Abiotic pollen and pollination: ecological, functional, and breakdown in self-incompatibility (regardless of the causal evolutionary perspectives. Plant Systematics and Evolution 222, 167-185 mechanism) is at all correlated with an increase in flowers Adams KL, Wendel JF (2005) Polyploidy and genome evolution in plants. Current Opinion in Plant Biology 8, 135-141 per inflorescence. Specifically, comparative phylogenetic Anderson E, Sax K (1936) Cytological monograph of the American species of analysis at the species level of Ranunculaceae may be re- Tradescantia. Botanical Gazette 97, 433-376 vealing of the order of trait evolution. Armbruster WS, Mulder CPH, Baldwin BG, Kalisz S, Wessa B, Nute H Within species, it appears that polyploids have larger (2002) Comparative analysis of late floral development and mating-system flowers than their diploid counterparts. Within species, in- evolution in tribe Collinsieae (Scrophulariaceae s.l.). American Journal of creased flower size has been observed to increase visitation Botany 89, 37-49 and reproductive success in a number of species (Bell 1985; Baack EJ (2005) Ecological factors influencing tetraploid, establishment in Connor 1997; Elle and Carney 2003). Our data on Plantago, snow buttercups (Ranunculus adoneus, Ranunculaceae): minority cytotype a wind-pollinated genus, indicate that increases in floral exclusion and barriers to triploid formation. American Journal of Botany 92, 1827-1835 size can occur regardless of the functional significance of Baker HG (1959) Reproductive methods as factors in speciation in flowering increased floral size. Although the gigas effect does seem to plants. Quantitative Biology 24, 177-191 exist in many polyploid species, it is not a universal feature Bali PN, Tandon SL (1959) Morphological and cytological studies of the of all polyploidization events. Our data indicate that associ- induced polyploids in Alyssum maritmum Lam. Genetica 30, 129-139 ations between polyploidy and larger flowers are only pre- Barrett SCH, Eckert CG (1990) Current issues in plant reproductive ecology. sent among autopolyploids and future efforts should con- Israel Journal of Botany 39, 5-12 centrate on why this might be the case. An important factor Bassett IJ, Crompton CW (1968) Pollen morphology and chromosome num- may be that the polyploids and diploids being compared are bers of the family Plantaginaceae in North America. Canadian Journal of separated by fewer generations in the autopolyploids for Botany - Revue Canadienne de Botanique 46, 349-361 which we have data, as the gigas effect has been seen to Bateman AJ (1952) Self-incompatibility systems in angiosperms. I. Theory. Heredity 6, 285-310 disappear within a few generations when there are selection Bawa KS (1990) Plant-pollinator interactions in tropical rain forests. Annual pressures to reduce floral size. Truly, there was surprisingly Review of Ecology and Systematics 21, 399-422 little information regarding flower size and polyploidy and Bell G (1985) On the function of flowers. Proceedings of the Royal Society of a careful assessment of the degree of publication bias is re- London 224, 223-265 quired before concrete conclusions can be made. Finally, Bennett JH (1976) Expectations for inbreeding depression on self-fertilization there is a decided lack of information on how the gigas ef- of tetraploids. Biometrics 32, 449-452 fect would manifest itself mechanistically. Developmental Bernadello G, Anderson GJ, Stuessy TF, Crawford DJ (2001) A survery of genetic studies indicate that the duplication of regulatory floral traits, breeding systems, floral visitors, and pollination systems of the angiosperms of the Juan Fernandez Islands (Chile). Botanical Review 67, MADS-box genes plays an important role in floral diver- 255-308 sification, yet whether polyploidy is responsible for these Brochmann C (1993) Reproductive strategies of diploid and polyploid popula- duplications is unknown (Kramer et al. 1998). tions of arctic Draba (Brassicaceae). Plant Systematics and Evolution 185, Polyploidy seems to have little association with other 55-83 floral traits. We necessarily divided Ranunculaceae into Brochmann C, Brysting AK, Alsos IG, Borgen L, Grundt HH, Scheen AC, coarse groupings based on symmetry and floral color and Elven R (2004) Polyploidy in arctic plants. Biological Journal of the Linnean found little indication that polyploidy has any effect on ei- Society 82, 521-536 ther trait, and therefore little evidence that polyploid esta- Brunet J, Eckert CG (1998) Effects of floral morphology and display on out- blishment is greatly facilitated when it arises within species crossing in Blue Columbine, Aquilegia caerulea (Ranunculaceae). Functional Ecology 12, 596-606 employing flower-constant pollinators (bumblebees, birds). Buide ML (2006) Pollination ecology of Silene acutifolia (Caryophyllaceae): Although inflorescence architecture has been posited as im- Floral traits variation and pollinator attraction. Annals of Botany 97, 289-297 portant in determining pollinator identity and behavior Busbice TH, Wilsie CP (1966) Inbreeding depression and heterosis in auto- (Harder et al. 2004), and thus could influence whether tetraploids with application to Medicago sativa. Euphytica 15, 52-67 polyploidy could establish, too little variation in inflores- Buxton BH, Darlington CD (1931) Behavior of a new species, Digitalis merto- cence type exists in either Rosaceae or Ranunculaceae for nensis. Nature 127, 94 us to perform a meaningful test. Therefore, relationships Carr GD (1978) Chromosome numbers of Hawaiian native plants and the sig- between inflorescence architecture (cyme, raceme, umbel) nificance of cytology in selected taxa. American Journal of Botany 65, 236- and polyploidy remains an uncharted avenue of future re- 242 search. Cohen D, Dukas R (1990) The optimal number of female flowers and the fruits-to-flowers ratio in plants under pollen and resource limitation. Amer- These newfound correlates between floral display and ican Naturalist 135, 218-241 polyploidy influence the interpretation of traditionally-re- Connor J (1997) Floral evolution in wild radish: the roles of pollinators, natural cognized correlates of polyploidy. Pollen delivery challen- selection, and genetic correlations among traits. International Journal of ges have long been recognized as a potential barrier for Plant Science 158, S108-S120 tetraploid establishment and this led to hypothesized cor- Crane MB, Lewis D (1942) Genetical studies in pears III. Incompatibility and relations between self-compatibility, perenniality, and poly- sterility. Journal of Genetics 43, 31-43 ploidy (Stebbins 1971). Yet despite the nearly 40 years of Cruden RW (2000) Pollen grains: why so many? Plant Systematics and Evolu- conjecture regarding how pollen delivery could change tion 222, 143-165 Cruden RW, Lyon DL (1985) Patterns of biomass allocation to male and upon polyploidization, we have remarkably few compar- female functions in plants with different mating systems. Oecologia 66, 299- ative or experimental examinations to base our theories on. 306 The need is readily apparent and we implore more empir- de Bodt S, Maere S, van de Peer Y (2005) Genome duplication and the origin ical pollination biologists to examine ploidy as a factor in of angiosperms. Trends in Ecology and Evolution 20, 591-597 their work. de Lange PJ, Murray BG (2003) Chromosome numbers of Norfolk Island endemic plants. Australian Journal of Botany 51, 211-215

Eckert CG (2000) Contributions of autogamy and geitonogamy to self-fertili- Lewis D (1947) Competition and dominance of incompatibility alleles in dip- zation in a mass-flowering, clonal plant. Ecology 81, 532-542 loid pollen. Heredity 1, 85-108 Elberling H, Olesen JM (1999) The structure of a high latitude plant-flower Liebst B, Schneller J (2005) How selfing and intra- and interspecific crossing visitor system: the dominance of flies. Ecography 22, 314-323 influence seed set, morphology, and ploidy level in Euphrasia: an experi- Elle E, Carney R (2003) Reproductive assurance varies with flower size in mental study of species occurring in the Alps of Switzerland. Plant Syste- Collinsia parviflora (Scrophulariaceae). American Journal of Botany 90, matics and Evolution 255, 193-214 888-896 Love A, Love D (1943) The significance of differences in the distribution of Emsweller SL, Ruttle ML (1941) Induced polyploidy in floriculture. Ameri- diploids and polyploids. Hereditas 29, 145-163 can Naturalist 75, 310-328 Lumaret R, Guillerm JL, Delay J, Ait Lhaj Loutfi A, Jay M (1987) Poly- French GC, Ennos RA, Silverside AJ Hollingsworth PM (2005) The rela- ploidy and habitat differentiation in Dactylis glomerata L. from Galicia tionship between flower size, inbreeding coefficient and inferred selfing rate (Spain). Oecologia 73, 436-446 in British Euphrasia species. Heredity 94, 44-51 Mable BK (2004) Polyploidy and self-compatibility: is there an association? Garbutt K, Bazzaz FA (1983) Leaf demography, flower production, and bio- New Phytologist 162, 803-811 mass of diploid and tetraploid populations of Phlox drummondii Hook. on a Macdonald SE, Chinnappa CC (1988) Patterns of variation in the Stellaria soil moisture gradient. New Phytologist 93, 129-141 longipes complex: effects of polyploidy and natural selection. American Jour- Gauthier P, Lumaret R, Bedecarrats A (1998) Genetic variation and gene nal of Botany 75, 1191-1200 flow in alpine diploid and tetraploid populations of Lotus (L. alpinus (D.C.) Medail F, Ziman S, Boscaiu M, Riera J, Lambrou M, Vela E, Dutton B, Schleicher/L. corniculatus L.). I. Insights from morphological and allozyme Ehrendorfer F (2002) Comparative analysis of biological and ecological dif- markers. Heredity 80, 683-693 ferentiation of Anemone palmata L. (Ranunculaceae) in the western Mediter- Grant V (1971) Plant Speciation, Columbia Press, New York, 563 pp ranean (France and Spain): an assessment of rarity and population persis- Guggisberg A, Mansion G, Kelso S, Conti E (2006) Evolution of biogeogra- tence. Biological Journal of the Linnean Society 140, 95-114 phic patterns, ploidy levels, and breeding systems in a diploid-polyploid spe- Miller JM, Chambers KL (1993) Nomenclature changes and new taxa in Clay- cies complex of Primula. New Phytologist 171, 617-632 tonia (Portulacaceae) in Western North America. Novon 3, 268-273 Hair JB (1966) Biosystematics of the New Zealand flora 1945-1964. New Zea- Miller JS, Venable DL (2000) Polyploidy and the evolution of gender dimor- land Journal of Botany 4, 559-595 phism in plants. Science 289, 2335-2338 Harder LD, Barrett SCH (1995) Mating cost of large floral displays in herma- Molau U (1993) Relationships between flowering phenology and life-history phrodite plants. Nature 373, 512-515 strategies in tundra plants. Arctic and Alpine Research 25, 391-402 Harder LD, Johnson SD (2005) Adaptive plasticity of floral display size in Morawetz W (1986) Remarks on karyological differentiation patters in tropical animal-pollinated plants. Proceedings of the Royal Society B - Biological woody plants. Plant Systematics and Evolution 152, 49-100 Sciences 272, 2651-2657 Newstrom L, Robertson A (2005) Progress in understanding pollination system Harder LD, Jordan CY, Gross WE, Routley MB (2004) Beyond floricen- in New Zealand. New Zealand Journal of Botany 43, 1-59 trism: the pollination function of inflorescences. Plant Species Biology 19, Nuismer SL, Cunningham BM (2005) Selection for phenotypic divergence 137-148 between diploid and autotetraploid Heuchera grossulariifolia. Evolution 59, Haskell G (1952) Polyploidy, ecology, and the British flora. Journal of Ecology 1928-1935 40, 265-282 Olesen JM, Eskildsen LI, Ventkatasamy S (2002) Invasion of pollination net- Haufler CH (1987) Electrophoresis is modifying our concepts of Evolution in works on oceanic islands: importance of invader complexes and endemic homosporous Pteridophytes. American Journal of Botany 74, 953-966 super generalists. Diversity and Distributions 8, 181-192 Hecht A (1944) Induced tetraploids of a self sterile Oenothera. Genetics 29, Olesen JM, Jordano P (2002) Geographic patterns in plant-pollinator mutua- 69-74 listic networks. Ecology 83, 2416-2424 Hilu KW (2004) Phylogenetics and chromosomal evolution in the Poaceae Osborn TC (2004) The contribution of polyploidy to variation in Brassica spe- (grasses). Australian Journal of Botany 52, 13-22 cies. Physiologia Plantarum 121, 531-536 Husband BC, Sabara HA (2004) Reproductive isolation between autotetra- Otto SP, Whitton J (2000) Polyploid incidence and evolution. Annual Review ploids and their diploid progenitors in fireweed, Chamerion angustifolium of Genetics 34, 401-437 (Onagraceae). New Phytologist 161, 703-713 Ownbey M (1950) Natural hybridization and amphiploidy in the genus Trago- Husband BC, Schemske DW (2000) Ecological mechanisms of reproductive pogon. American Journal of Botany 37, 487-499 isolation between diploid and tetraploid Chamerion angustifolium (Onagra- Pannell JR, Obbard DJ, Buggs RJA (2004) Polyploidy and sexual system: ceae). Journal of Ecology 88, 689-701 what can we learn from Mercurialis annua? Biological Journal of the Lin- Jacquemart AL, Thompson JD (1996) Floral and pollination biology of three nean Society 82, 547-560 sympatric Vaccinum (Ericaceae) species in the Upper Ardennes, Belgium. Perez F, Arroyo MTK, Medel R, Hershkovitz MA (2006) Ancestral recons- Canadian Journal of Botany 74, 210-221 truction of flower morphology and pollination systems in Schizanthus (Sola- Johnson AW, Packer JG (1965) Polyploidy and Environment in Arctic Alaska. naceae). American Journal of Botany 93, 1029-1038 Science 148, 237 Petit C, Lesbros P, Ge X, Thompson JD (1997) Variation in flowering Johnston MO, Schoen DJ (1996) Correlated evolution of self-fertilization and phenology and selfing rate across a contact zone between diploid and inbreeding depression: an experimental study of nine populations of Am- tetraploid Arrhenatherum elatius (Poaceae). Heredity 79, 31-40 sinckia (Boraginaceae). Evolution 50, 1478-1491 Petit C, Thompson JD (1999) Species diversity and ecological range in relation Kalkman (2004) Rosaceae. In: Kubitzki K (Ed) The Families and Genera of to ploidy level in the flora of the Pyrenees. Evolutionary Ecology 13, 45-66 Vascular Plants. VI. Flowering Plants. Dicotyledons. Celastrales, Oxalidales, Primack RB (1978) Evolutionary aspects of wind pollination in genus Plan- Rosales, Cornales, Ericales, Springer, Berlin, pp 343-386 tago (Plantaginaceae). New Phytologist 81, 449-458 Karron JD, Mitchell RJ, Holmquist KG, Bell JM, Funk B (2004) The influ- Rademaker MCJ, de Jong TJ, van der Meijden E (1999) Selfing rates in ence of floral display size on selfing rates in Mimulus ringens. Heredity 92, natural populations of Echium vulgare: a combined empirical and model ap- 242-248 proach. Functional Ecology 13, 828-837 Kennedy BF, Sabara HA, Haydon D, Husband BC (2006) Pollinator-medi- Rahn K (1996) A phylogenetic study of the Plantaginaceae. Botanical Journal ated assortative mating in mixed ploidy populations of Chamerion angusti- of the Linnean Society 120, 145-198 folium (Onagraceae). Oecologia 150, 398-408 Ramsey J, Schemske DW (1998) Pathways, mechanisms, and rates of poly- Knight TM, Steets JA, Vamosi JC, Mazer SJ, Burd M, Campbell DR, Du- ploidy formation in flowering plants. Annual Review of Ecology and Syste- dash MR, Johnston MO, Mitchell RJ, Ashman T-L (2005) Pollen limita- matics 29, 467-501 tion of plant reproduction: pattern and process. Annual Review of Ecology Ramsey J, Schemske DW (2002) Neopolyploidy in flowering plants. Annual Evolution and Systematics 36, 467-497 Review of Ecology and Systematics 33, 589-639 Kramer EM, Dorit RL, Irish VF (1998) Molecular evolution of genes con- Randolph LF (1935) Cytogenetics of tetraploid maize. Journal of Agricultural trolling petal and stamen development: duplication and divergence within the Research 50, 591-605 APETALA3 and PISTILLATA MADS-Box gene lineages. Genetics 149, Rausch JH, Morgan MT (2005) The effect of self-fertilization, inbreeding de- 765-783 pression, and population size on autopolyploid establishment. Evolution 59, Lande R, Schemske DW (1985) The evolution of self-fertilization and in- 1867-1875 breeding depression in plants. I. Genetic models. Evolution 39, 24-40 Rausch JH, Morgan MT (2006) Formulation of competition between sexual Leclerc-Potvin C, Ritland K (1994) Modes of self-fertilization in Mimulus- and selfing forms: A response to Yamauchi. Evolution 60, 1954-1955 Guttatus (Scrophulariaceae) - a field experiment. American Journal of Bot- Regal PJ (1982) Pollination by wind and animals: ecology of geographic pat- any 81, 199-205 terns. Annual Review of Ecology and Systematics 13, 497-524 Levan A (1941) The cytology of the species hybrid Allium cepa x fistulosum Richards AJ (1997) Plant Breeding Systems, Chapman Hall, London, 529 pp and its polyploid derivatives. Hereditas 27, 253-272 Rosquist G, Prentice HC (2001) Morphological variation in Scandinavian Levin DA (1975) Minority cytotype exclusion in local plant populations. Taxon populations of the diploid-tetraploid species pair Anthericum ramosum and 24, 35-43 Anthericum liliago (Anthericaceae). Canadian Journal of Botany 79, 850-860 Levin DA (1983) Polyploidy and novelty in flowering plants. American Nat- Rothera SL, Davy AJ (1986) Polyploidy and habitat differentiation in Des- uralist 122, 1-25 champsia cespitosa. New Phytologist 102, 449-467

8 Pollination ecology of polyploidy. Vamosi et al.

Ruiz Rejon C, Blanca G, Cueto M, Lozano R, Ruiz Rejon M (1990) Aspho- plants: origins, establishment and persistence. Trends in Ecology and Evol- delus tenuifolius and A. fistulosus (Liliaceae) are morphologically, geneti- ution 7, 302-307 cally, and biologically different species. Plant Systematics and Evolution 169, Thompson JD, Nuismer SL, Merg K (2004) Plant polyploidy and the evol- 1-12 utionary ecology of plant/animal interactions. Biological Journal of the Lin- Sahley CT (1996) Bat and hummingbird pollination of an autotetraploid co- nean Society 82, 511-519 lumnar cactus, Weberbauerocereus weberbaueri (Cactaceae). American Jour- Thompson SL, Whitton J (2006) Patterns of recurrent evolution and geog- nal of Botany 83, 1329-1336 raphic parthenogenesis within apomictic polyploid Easter daises (Townsendia Sargent RD (2004) Floral symmetry affects speciation rates in angiosperms. hookeri). Molecular Ecology 15, 3389-3400 Proceedings of the Royal Society of London B 271, 603-608 Totland O (2001) Environment-dependent pollen limitation and selection on Sato H, Yahara T (1999) Trade-offs between flower number and investment to floral traits in an alpine species. Ecology 82, 2233-2244 a flower in selfing and outcrossing varieties of Impatiens hypohylla. Amer- Valido A, Dupont YL, Olesen JM (2004) Bird-flower interactions in the Mac- ican Journal of Botany 86, 1699-1707 ronesian islands. Journal of Biogeography 31, 1945-1953 Schoen DJ, Morgan MT, Bataillon T (1996) How does self-pollination Vamosi JC, Dickinson TA (2006) Polyploidy and diversification: A phylogen- evolve? Inferences from floral ecology and molecular genetic variation. Phi- etic investigation in Rosaceae. International Journal of Plant Sciences 167, losophical Transactions of the Royal Society of London Series B - Biological 346-355 Sciences 351, 1281-1290 Vamosi JC, Otto SP (2002) When looks can kill: the evolution of sexually- Segraves KA, Thompson JN (1999) Plant polyploidy and pollination: floral dimorphic floral display and the extinction of dioecious plants. Proceedings traits and insect visits to diploid and tetraploid Heuchera grossulariifolia. of the Royal Society of London B 269, 1187-1194 Evolution 53, 1114-1127 van der Niet T, Johnson SD, Linder HP (2006) Macroevolutionary data sug- Smith AP, Young TP (1987) Tropical alpine plant ecology. Annual Review of gest a role for reinforcement in pollination system shifts. Evolution 60, 1596- Ecology and Systematics 18, 137-158 1601 Smith H (1946) Sedum pulchellum: a physiological and morphological compa- van Huylenbroeck JM, de Riek J, de Loose M (2000) Genetic relationships rison of diploid, tetraploid, and hexaploid races. Bulletin of the Torrey Bot- among Hibiscus syriacus, Hibiscus sinosyriacus and Hibiscus paramutabilis anical Society 73, 495-541 revealed by AFLP, morphology, and ploidy analysis. Genetic Resources and Snell R, Aarssen LW (2005) Life history traits in selfing versus outcrossing Crop Evolution 47, 335-343 annuals: exploring the 'time-limitation' hypothesis for the fitness benefit of Wallace LE (2004) A comparison of genetic variation and structure in the allo- self-pollination. BMC Ecology 5, 2 polyploid Platanthera huronensis and its diploid progenitors, Platanthera Soltis DE, Soltis PS (1993) Molecular data and the dynamic nature of poly- aquilonis and Platanthera dilatata (Orchidaceae). Canadian Journal of Bot- ploidy. Critical Reviews in Plant Sciences 12, 243-273 any 82, 244-252 Stebbins GL (1957) Self fertilization and population variability in the higher Webb CJ, Lloyd DG (1986) The avoidance of interference between the pres- plants. American Naturalist 91, 337-54 entation of pollen and stigmas in angiosperms. 2. Herkogamy. New Zealand Stebbins GL (1971) Chromosomal Evolution in Higher Plants, Edward Arnold, Journal of Botany 24, 163-178 London, 216 pp Wendel JF (2000) Genome evolution in polyploids. Plant Molecular Biology Stebbins GL (1980) Rarity of plant species - synthetic viewpoint. Rhodora 82, 42, 225-249 77-86 Whitton J (2004) One down and thousands to go - dissecting polyploid speci- Tamura M (2004) Ranunculaceae. In: Kubitzki K, Rohwer JG, Bittrich V (Eds) ation. New Phytologist 161, 610-612 The Families and Genera of Vascular Plants. II. Flowering Plants. Dicoty- Williams W (1951) Genetics of incompatibility in alsike clover, Trifolium hyb- ledons. Magnolid, Hamamelid and Caryophyllid Families, Springer, Berlin, ridum. Heredity 5, 51-73 pp 563-585 Worley AC, Baker AM, Thompson JD, Barrett SCH (2000) Floral display in Tate JA, Simpson BB (2004) Breeding system evolution in Tarasa (Malva- Narcissus: Variation in flower size and number at the species, population, and ceae) and selection for reduced pollen grain size in the polyploid species. individual levels. International Journal of Plant Sciences 161, 69-79 American Journal of Botany 91, 207-213 Yamauchi A (2006) Formulation of competition between sexual and selfing Taylor NL, Smith RR (1979) Red clover breeding and genetics. Advanced forms. Evolution 60, 1952-1953 Agronomy 31, 125-154 Yeung K, Miller JS, Savage AE, Husband BC, Igic B, Kohn JR (2005) Asso- Thompson JD, Cunningham BM, Segraves KA, Althoff DM, Wagner D ciation of ploidy and sexual system in Lycium californicum (Solanaceae). (1997) Plant polyploidy and insect/plant interactions. American Naturalist Evolution 59, 2048-2055 150, 730-743 Yumoto T (1986) The ecological pollination syndromes of insect-pollinated Thompson JD, Lumaret R (1992) The evolutionary dynamics of polyploid plants in an alpine meadow. Ecological Research 1, 83-95